Skip to main content
Lab
Interactive Demo
Data Engineering

Revenue Leak Investigation

An autonomous AI agent fleet diagnosed EUR 2.1M in lost revenue from a payment gateway failure — zero SQL queries written. Step through the full investigation below.

0.0M

Revenue Impact

lost over 47 undetected days

< 0 min

Detection Time

automated agent investigation

0

SQL Queries Written

fully agent-driven analysis

0

Agent Delegations

LangGraph multi-agent fleet

Agent Investigation

Watch the autonomous agent fleet diagnose the revenue leak in real-time. Click Launch Investigation to begin, or use the phase tabs to jump to any step.

agent-investigation.log

4 autonomous agents · 33 log entries · < 15 min

1/4
Agent Fleet

Click each agent to see its role in the investigation and key finding.

Conversion Rate Explorer

Interactive chart showing checkout-to-purchase conversion over 106 days. Toggle between mobile and desktop to see how the anomaly only affected mobile traffic.

Mobile conversion
Anomaly window
Revenue Impact Analysis

Drag the slider to see how earlier detection would have reduced losses. At €44,681/day, every day matters.

Adyen 3DS2 iframe redirect fails on mobile Safari€1,200,000
SameSite=Strict cookie policy drops session on Chrome Android€620,000
Empty payment_token from gateway — silent transaction failure€180,000
What-If AnalysisDetection on day 47 of 47
Day 1
Day 47

Actual Loss

€2.1M

Revenue Saved

€0

At €44,681/day, every day of delayed detection costs the equivalent of a senior engineer's annual salary.

Root Cause Deep Dive

Click each finding to expand the evidence chain and remediation details.

Medallion Architecture

Click each layer to explore the Bronze → Silver → Gold pipeline that powers the investigation.

Stack

Apache Spark 3.5

Distributed data processing engine

Delta Lake

ACID transactions on the data lake

MinIO

S3-compatible object storage

LangGraph

Agent state machine orchestration

FastAPI

Agent API gateway & endpoints

Qwen 2.5

Local LLM for agent reasoning

Airflow

Pipeline scheduling & monitoring

PostgreSQL

Metadata & session storage

Docker

Container orchestration

Prometheus

Metrics collection & alerting

This investigation demonstrates how autonomous AI agents eliminate the hypothesis-query-iterate cycle — from anomaly detection through multi-agent root cause analysis to automated remediation deployment.

Book a 30-Min Call